[CRV 2022, Oral] Occlusion-Aware Self-Supervised Stereo Matching with Confidence Guided Raw Disparity Fusion - View it on GitHub